γ-Aminobutyric acid and glutamate dysregulation in the dorsolateral prefrontal cortex of adolescents with first-episode major depressive disorder and the modulatory effects of repetitive transcranial magnetic stimulation

Abstract

Background

Major depressive disorder (MDD) is associated with dysregulation of γ-aminobutyric acid (GABA) and glutamate(Glu) neurotransmission in the prefrontal cortex. Proton magnetic resonance spectroscopy (1H-MRS) enables non-invasive in vivo quantification of GABA and Glx(glutamate + glutamine) levels. This study investigated neurochemical characteristics of the bilateral dorsolateral prefrontal cortex (DLPFC) in first-episode adolescent MDD (FEA-MDD) and repetitive transcranial magnetic stimulation (rTMS)’s impact on these changes.

Methods

42 drug-naïve FEA-MDD patients underwent bilateral DLPFC MRS scans before and after rTMS, with 42 healthy controls (HCs) as baseline. All participants were right-handed. The Mescher–Garwood point-resolved spectroscopy (MEGA-PRESS) protocol detected GABA+ (GABA plus macromolecules and high carnosine) and Glx levels, processed via Gannet software.

Results

FEA-MDD patients exhibited significantly lower GABA+ and higher Glx levels in the left DLPFC than HCs; in the right DLPFC, no significant difference in GABA+ levels was observed, though Glx levels were elevated. After rTMS treatment, GABA+ levels in the left DLPFC increased significantly, whereas Glx levels showed a non-significant decreasing trend. Additionally, HCs had no hemispheric differences, while in FEA-MDD, the left DLPFC showed lower GABA+ and Glx levels compared to the right. We also found that in the left DLPFC, baseline GABA+ levels were negatively correlated with Hamilton Depression Scale (HAMD) scores; Glx levels showed positive correlations with scores on the Ruminative Response Scale (RRS), Self-Rating Depression Scale (SDS), and Self-Esteem Scale (SES).

Conclusions

FEA-MDD involves prefrontal GABA+/Glx dysregulation, and rTMS may aid in restoring neurotransmitter balance within the DLPFC. This study adds to the expanding body of evidence supporting the application of targeted neurochemical modulation in the treatment of FEA-MDD, while also providing insights into potential intervention mechanisms.

Introduction

Major depressive disorder (MDD) in adolescents is a significant public health concern, affecting approximately 10–20% of youth globally (McGrath et al., 2023). Adolescent MDD is characterized by persistent feelings of sadness, loss of interest in activities, and a range of cognitive and physical symptoms. Adolescence is a period of rapid brain development, particularly in the systems responsible for emotional and cognitive regulation. Depression during this critical developmental stage can disrupt normal neurobiological maturation, causing long-term impairments in academic, social, and emotional functioning (Xu et al., 2024). Furthermore, adolescent MDD is often more severe and treatment-resistant than adult-onset depression (Malinina et al., 2023), and early-onset depression is associated with a higher risk of recurrent episodes and the development of other mental health disorders in adulthood (Malhi and Mann, 2018; Xu et al., 2024). As such, understanding the underlying neurobiological mechanisms of adolescent MDD is crucial for developing more effective interventions.

The pathophysiology of MDD is multifactorial, involving complex interactions between genetic, environmental, and neurobiological factors. Neuroimaging studies have consistently demonstrated structural and functional alterations in key brain regions implicated in mood regulation, including the prefrontal cortex (PFC), amygdala, and hippocampus (Levitt et al., 2019; Yun and Kim, 2021; Jacob et al., 2022). Among these regions, the dorsolateral PFC (DLPFC) has garnered particular attention due to its critical role in higher-order cognitive functions, including decision-making, emotional regulation, and executive control. Dysregulation within the DLPFC is hypothesized to contribute to the cognitive and emotional disturbances observed in MDD (Cutler et al., 2023). Furthermore, a growing body of evidence indicates that imbalances in key neurotransmitter systems, particularly γ-aminobutyric acid (GABA) and (Glx), play a central role in the pathophysiology of depression (Su et al., 2020; Jiao et al., 2024).

GABA and Glx are the principal inhibitory and excitatory neurotransmitters in the central nervous system, respectively (Lüscher and Möhler, 2019; Adell, 2020). Both systems have been implicated in MDD, with documented effects on brain regions involved in mood regulation and cognitive control, such as the PFC, hippocampus, and amygdala (Sanacora et al., 1999, 2012; Geurts et al., 2004). Deficits in GABAergic signaling are well-established in MDD and are thought to impair the brain’s capacity to regulate excessive neural activity, particularly within regions such as the DLPFC (Sanacora et al., 2002; Miller et al., 2009). Furthermore, reduced GABA levels in the PFC have been linked to depressive symptoms, including impaired emotional regulation, diminished cognitive control, and heightened stress reactivity (Levitt et al., 2019). Conversely, Glx, the principal excitatory neurotransmitter, has been proposed to play a role in depression through its excitotoxic effects. Elevated Glu levels have been observed in multiple MDD studies, and such increases are hypothesized to damage and impair synaptic plasticity (Arnone et al., 2015; Abdallah et al., 2017). The Glu hypothesis of depression posits that chronic overactivation of Glx receptors, particularly N-methyl-d-aspartate (NMDA) receptors, may induce excitotoxicity and neuronal loss. This, in turn, impairs cognitive flexibility, emotional regulation, and stress responses, which represent core deficits in MDD (Su et al., 2020). Thus, maintaining a proper balance between GABAergic and glutamatergic systems is critical for emotional stability and cognitive function.

Despite significant advances in understanding the neurobiology of MDD, effective treatments for this condition remain limited. Conventional antidepressant therapies, including pharmacological and psychotherapeutic approaches, often fail to address the underlying neurobiological dysfunctions in adolescents, and are associated with delayed onset of action, adverse effects, and poor response in a substantial proportion of patients (Levitt et al., 2019; Godfrey et al., 2021). Pharmacological agents, particularly selective serotonin reuptake inhibitors (SSRIs), target monoaminergic systems (serotonin, norepinephrine, and dopamine) but typically require several weeks to exert effects, with approximately one-third of patients failing to achieve adequate responses (Kanekar et al., 2018). This has spurred growing interest in alternative non-pharmacological treatments such as repetitive transcranial magnetic stimulation (rTMS), which offers a promising means of modulating the neural circuits implicated in depression without inducing the systemic adverse reactions associated with medications.

rTMS is a non-invasive neuromodulation technique that uses magnetic pulses to induce electrical currents in specific brain regions, thereby regulating neural activity and alleviating depressive symptoms (Croarkin et al., 2011; McClintock et al., 2018; Breda and Freire, 2024). The DLPFC is a primary target of rTMS, given its critical role in mood regulation and executive functioning. Clinical studies have demonstrated that rTMS can significantly improve depressive symptoms by enhancing neuronal activity in the DLPFC, which in turn modulates the neural circuits underlying mood regulation (Richieri et al., 2017; Lu et al., 2023). Growing evidence also indicates that rTMS can modulate the balance between excitatory and inhibitory neurotransmission, potentially augmenting GABAergic activity in the DLPFC to counteract the GABA deficits observed in MDD (Levitt et al., 2019; Godfrey et al., 2021). Furthermore, post-rTMS alterations in frontal Glu and Glx concentrations have been proposed as potential biomarkers for predicting clinical outcomes (Bhattacharyya et al., 2021; Godfrey et al., 2021; Gonsalves et al., 2022).

While the clinical efficacy of rTMS in adult populations is well-established, its effects in adolescents with MDD remain underexplored. Adolescence is a period of profound brain maturation, and neurobiological responses to rTMS may differ between adolescents and adults due to age-related variations in brain structure, function, and neurochemical systems (Croarkin et al., 2021; Hett et al., 2021). Thus, understanding how rTMS modulates neurochemical profiles in adolescents is critical for refining treatment protocols and identifying potential biomarkers for treatment response.

In this study, we hypothesize that first-episode adolescent major depressive disorder (FEA-MDD) exhibits a distinct imbalance between GABA and Glx in the DLPFC, and that the therapeutic effects of rTMS are mediated by the normalization of this imbalance. Utilizing ¹H-MRS, this study aims to: (i) quantify baseline GABA+ and Glx levels in the bilateral DLPFC of FEA-MDD patients compared to healthy controls (HCs); (ii) evaluate the neurochemical effects of a 2-week rTMS treatment protocol; and (iii) explore correlations between neurotransmitter levels and clinical symptom severity. Our goal is to elucidate the neurochemical mechanisms of rTMS and identify potential biomarkers for treatment response in adolescent depression.

Materials and methods

Participants

A total of 42 newly diagnosed FEA-MDD patients meeting the criteria of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) were recruited from the Mental Health Branch of Lanzhou University Second Hospital. Additionally, 42 HCs matched for age, sex, and education were recruited from the community via hospital advertisements. The specific inclusion criteria for FEA-MDD were as follows (Lu et al., 2023): (i) age 13–21 years; (ii) of Han Chinese ethnicity; (iii) diagnosed with MDD as inpatients or outpatients based on DSM-IV criteria, with a total score ≥20 on the 24-item Hamilton Depression Rating Scale (HAMD-24); (iv) no obvious organic lesions in the brain parenchyma on the T2-weighted imaging (T2WI) sequence; and (v) experiencing their first depressive episode, with no prior treatment with antipsychotic medications or psychotherapy.

Exclusion criteria for the patient group were as follows (Bhattacharyya et al., 2021): (i) a history of any antidepressant treatment prior to the diagnosis depression; (ii) presence of other neurological or psychiatric illness; (iii) presence of significant acute suicidal ideation or risk of suicidal behavior; (iv) acute physical illness that could affect the scans; (v) claustrophobia or metal implants that made the patient unsuitable for magnetic resonance scans; (vi) failure to abstain from alcohol, tea, and caffeine for at least 24 h prior to the magnetic resonance imaging (MRI) scan; or (vii) contraindications to rTMS treatment (e.g. cardiac pacemaker, history of epilepsy, or metals or magnetic implants). In addition, HCs were screened for neurological or psychiatric disorders (see Table 1 for details).

Table 1.

Demographic and neuropsychological characteristics of the participants.

				HC vs. FEA-MDD		FEA-MDD vs. Post-rTMS
Characteristic	HC, n = 42	FEA-MDD, n = 42	Post-rTMS, n = 42	T	P	t	P
Sex (M/F)	20/22	15/27	−	1.48	>0.999	−	−
Age (years)	18.61 ± 3.52	16.14 ± 1.82	−	4.04	<0.0001	-	-
Education (years)	12.11 ± 3.40	11.19 ± 3.45	−	1.24	0.218	−	−
MoCA	26.14 ± 1.81	26.19 ± 0.96	26.67 ± 1.24	−0.23	0.81	−1.95	0.057
HAMD	3.40 ± 2.11	25.74 ± 7.80	17.10 ± 7.42	−17.91	<0.0001	5.2	<0.0001
HAMA	3.00 ± 1.81	22.93 ± 8.26	16.88 ± 6.82	−15.28	<0.0001	9.71	0.0004
RRS	29.76 ± 3.62	60.33 ± 11.11	44.50 ± 10.97	−16.95	<0.0001	6.57	<0.0001
SES	27.28 ± 2.63	28.79 ± 5.02	27.12 ± 4.37	−1.72	0.092	1.62	0.11
SHAPS	45.04 ± 3.54	34.61 ± 7.78	28.14 ± 7.03	8.34	<0.0001	3.62	0.0005
SDS	34.17 ± 7.99	57.10 ± 7.11	45.50 ± 9.71	−13.89	<0.0001	6.24	<0.0001
SAS	30.95 ± 7.44	50.21 ± 8.60	40.40 ± 9.11	−10.98	<0.0001	5.08	<0.0001

Note: Data are presented as mean ± SD or counts. n, number of participants; HC, healthy controls; FEA-MDD, first episode adolescent major depressive disorder; Post-rTMS, after 2 weeks of rTMS treatment; MoCA, Montreal Cognitive Assessment; HAMD, Hamilton Depression Rating Scale; HAMA, Hamilton Anxiety Scale; RRS, Ruminative Responses Scale; SES, Self-Esteem Scale; SHAPS, Snaith–Hamilton Pleasure Scale; SDS, Self-Rating Depression Scale; SAS, Self-Rating Anxiety Scale.

All participants were informed of the purpose and content of the experiment and voluntarily provided informed consent. This study was approved by the Medical Ethics Committee of Lanzhou University Second Hospital (2022A-395).

rTMS procedure

All FEA-MDD patients underwent rTMS targeting the DLPFC using a high-focus "8" coil. In accordance with transcranial magnetic stimulation treatment guidelines (McClintock et al., 2018), the rTMS parameters were as follows: the treatment was administered once daily for 10 consecutive days, with an intensity set at 100% of the motor threshold. The stimulation frequency was 10 Hz for 2 s, with a total of 1200 stimuli delivered over 10 min. The left DLPFC was scalp-localized using the "5 cm method" (Tsukuda et al., 2024). All patients were maintained on stable antidepressant regimens throughout the 2-week rTMS treatment period (see Supplementary Fig. S1 for a flow diagram of the inclusion process).

Clinical assessment

The severity of depression and anxiety was assessed using the HAMD-24 and the 14-item Hamilton Anxiety Rating Scale (HAMA) at two time points: (i) within 1–2 days before the first rTMS treatment session (baseline) and (ii) within 1–2 days after the final rTMS session (post-rTMS) (Hamilton, 1967).

Clinical response was evaluated primarilybased on the reduction in HAMD scores, calculated using the formula:

Response categories were defined as follows:

• Cured: reduction rate ≥75% or HAMD-17 score <7

• Markedly effective: 50% ≤ reduction rate < 74%

• Effective: 25% ≤ reduction rate < 49%

• Ineffective: reduction rate < 25%

The responders was calculated as the sum of the cure rate, markedly effectiveness rate, and effectiveness rate (Jiao et al., 2024).

Additionally, other clinical measures were administered to assess various symptoms, including the Montreal Cognitive Assessment (MoCA), Self-Esteem Scale (SES), Snaith–Hamilton Pleasure Scale (SHAPS), and Ruminative Responses Scale (RRS).

Imaging acquisition

All imaging protocols were performed on a 3T MR scanner (SIGNA Premier; GE Healthcare, Waukesha, WI, USA) using a 48-channel head coil. Magnetic resonance spectroscopy (MRS) data were acquired with the MEGA-PRESS sequence. Volumes of interest (VOIs) were positioned at the center of the left and right DLPFC. The orientation of the orthogonal plane (axial, sagittal, and coronal) was determined using high-resolution 3D T1-weighted imaging (T1WI) reconstruction, with exact locations illustrated in Fig. 1A. Voxel dimensions were 20 × 20 × 20 mm³. MEGA-PRESS MRS scanning parameters were as follows: repetition time/echo time (TR/TE) = 2000/68 ms, spectral width = 2000 Hz, number of excitations (NEX) = 8 per condition (edit-ON and edit-OFF), and the unilateral acquisition duration = 11 min 28 s. The scanning parameters for 3D-T1WI were: sagittal plane, TR/TE = 2632/3.0 ms, inversion time (TI) = 1000 ms, field of view (FOV) = 256 × 256 mm², slice thickness = 1 mm, slice gap = 0.5 mm, slices = 392, flip angle = 8°, matrix = 256 × 256, parallel acquisition factor = 2, bandwidth = 31.25 kHz, resolution = 1 × 1 × 1 mm, and the average acquisition duration = 6 min 8 s.

Figure 1.

Regions of interest and MRS fitting curves. (A) Location of sampling points in the left and right dorsolateral prefrontal cortex. (B) Difference spectrum and model fitting results. (C) Reference signals. (D) Segmentation of gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF).

MEGA-PRESS sequence generates two subspectra: one with the editing pulse (edit-ON) and one without (edit-OFF). The editing pulse was set to ON at 1.9 ppm and OFF at 7.46 ppm. Subtracting the edit-OFF spectrum from the edit-ON spectrum eliminated the total creatine (tCr) signal from the edited spectrum, enabling quantification of GABA+ levels (Mullins et al., 2014).

¹H-MRS data processing and quantification

MRS data were analyzed using the Gannet 3.1 software toolkit, an open-source GABA-MRS analysis tool developed in MATLAB 2022b (The MathWorks, Inc., Natick, MA, USA) with the Optimization and Statistics toolbox. The workflow consisted of three main steps. (i) Signal preprocessing: frequency and phase correction were applied to align free induction decay (FID) signals. Robust spectral correction algorithms were utilized to minimize artifacts from magnetic field inhomogeneity. (ii) Edited spectrum generation: spectra from edit-ON and edit-OFF scans were averaged and subtracted to isolate GABA+-specific signals (central frequency at 3.02 ppm). (iii) Metabolite quantification: the GABA+ peak at 3 ppm was fitted with a Gaussian function, and its intensity was quantified by the area under the curve. Results were expressed in institutional units (i.u.) and normalized to the unsuppressed water signal.

Data quality control: Spectral quality was evaluated using the fitting error (FitErr), defined as the ratio of the standard deviation of fitting residuals to the GABA+ signal amplitude. Given the lateral location of the DLPFC (adjacent to the skull) and the small VOI collected, both contributing to a low signal-to-noise ratio (SNR), a FitErr threshold of 20% was applied to retain valid data. While relaxing the threshold may introduce minor noise, it provides a practical quality screening solution for studies involving small VOIs. This threshold can be adjusted in future in vivo research based on cohort characteristics to optimize statistical power. Spectroscopic voxels were co-registered to individual 3D-T1WI anatomical images using SPM12 within the Gannet toolbox. Subsequently tissue segmentation was performed to distinguish gray matter, white matter, and cerebrospinal fluid (CSF). Given the minimal GABA contribution from CSF, a gray matter-dominant mask was applied for partial volume correction, ensuring precise quantification of GABA+ levels and reducing interference from non-neuronal tissues.

Statistical analysis

Statistical analyses were performed using SPSS 26.0 software. Continuous data were expressed as mean ± SD. The Kolmogorov–Smirnov method was used to test the normality of the data (e.g. age, education level), and all continuous variables showed a normal distribution. Independent samples t-tests were conducted to compare baseline differences in bilateral DLPFC GABA+ and Glx levels, as well as HAMD and HAMA scores and other clinical scales, between HC and FEA-MDD groups. Paired samples t-tests were used to evaluate within-group changes in these measures before and after rTMS treatment in the FEA-MDD group. Pearson correlation analysis was performed to examine the correlations between clinical scale scores and GABA+ and Glx levels. False discovery rate (FDR) correction was applied to all t-test results to control for multiple comparisons, ensuring the validity of statistical inferences. A post-hoc power analysis was conducted using G*Power 3.1 for the primary neurochemical outcome. Statistical significance was set at the threshold of P < 0.05.

Results

Demographic and clinical characteristics

No significant differences were observed between the HC and FEA-MDD groups in terms of sex, age, or education level. However, the two groups showed significant differences in HAMD, HAMA, SES, RRS, and SHAPS scores. Detailed statistics are presented in Table 1.

Baseline neurotransmitter levels in the DLPFC

At baseline, neurotransmitter analyses revealed significant differences between the FEA-MDD and HC groups. Specifically, the FEA-MDD group exhibited significantly lower GABA+ levels (P < 0.01) and higher Glx levels (P < 0.01) in the left DLPFC compared to HCs. In the right DLPFC, while the FEA-MDD group showed higher Glx levels (P < 0.01), no significant differences in GABA+ levels were observed between the two groups (P > 0.05). Detailed statistics are presented in Table 2 and Fig. 2.

Table 2.

Baseline MRS voxel segmentation percentages and metabolite levels in the DLPFC.

Characteristic	HC (n = 42)	FEA-MDD (n = 42)	t	P	Cohen's d
GABA+ (L)	1.48 ± 0.27	1.30 ± 0.21	3.46	0.0009	0.744
Glx (L)	4.90 ± 1.04	5.37 ± 0.69	−2.47	0.0153	0.530
Gray matter (L), %vol.	46.77 ± 6.04	44.36 ± 5.76	1.91	0.059	0.069
White matter (L), %vol.	37.12 ± 6.07	37.86 ± 6.58	−0.53	0.592	0.116
CSF (L), %vol.	15.24 ± 3.78	14.22 ± 3.85	1.23	0.221	0.261
GABA+ (R)	1.55 ± 0.37	1.59 ± 0.34	0.59	0.556	0.112
Glx (R)	4.45 ± 1.06	5.83 ± 1.20	−5.58	<0.0001	1.20
Gray matter (R), %vol.	44.23 ± 5.56	45.90 ± 5.08	−1.37	0.1749	0.313
White matter (R), %vol.	38.12 ± 4.78	37.96 ± 6.54	0.12	0.90	0.027
CSF (R), %vol.	14.36 ± 3.14	15.12 ± 3.74	−0.96	0.3381	0.220

Note: data are presented as mean ± SD. n, number of participants; HC, healthy controls; FEA-MDD, first episode adolescent major depressive disorder; DLPFC, dorsolateral prefrontal cortex; L, left; R, right; CSF, cerebrospinal fluid; Cohen’s d represents the standardized effect size for between-group differences.

Figure 2.

Baseline metabolite levels in the left and right DLPFC of the HC and FEA-MDD groups. (A) GABA+ levels in the left DLPFC were significantly lower in the FEA-MDD group than in HCs (P < 0.05). (B) Glx levels in the left DLPFC were significantly higher in adolescents with first-episode depression than in HCs (P < 0.05). (C) GABA+ levels in the right DLPFC showed no significant difference between the two groups. (D) Glx levels in the right DLPFC were significantly higher in adolescents with first-episode depression than in HCs (P < 0.05). *FDR‐adjusted P  < 0.05, ***FDR‐adjusted P  < 0.001, ****FDR‐adjusted P  < 0.0001. L, left; R, right.

Effects of rTMS treatment

After 2 weeks of rTMS treatment, the FEA-MDD group showed significant reductions in scores on the HAMD, HAMA, RRS, SHAPS, SDS, and SAS (see Table 1 for details). Of the 42 FEA-MDD participants, eight were classified as "cured," 17 as "markedly effective," and 10 as "effective," resulting in a total of 35 responders. Seven participants were classified as "ineffective." The average reduction in depression scores among the 35 responders was 60.8% from baseline to post-rTMS assessment.

Changes in neurotransmitter levels

We further analyzed neurotransmitter levels in the left and right DLPFC of FEA-MDD participants stratified by treatment response (see Supplementary Table S1 for details). In responders (the total effective cases), post-rTMS assessments revealed significant neurochemical changes in the left DLPFC: GABA+ levels increased significantly (P < 0.05), while Glx levels showed a decreasing trend (though not statistically significant, P = 0.188). These findings suggest that rTMS may help restore a more balanced neurochemical profile in the left DLPFC. No significant changes in GABA+ or Glx levels were observed in the right DLPFC following rTMS (Table 3; Fig. 3). The trajectories of neurotransmitter levels before and after rTMS treatment are presented in Supplementary Fig. S2. Additionally, comparisons of post-rTMS neurotransmitter levels with those of HCs are provided in Supplementary Table S2 and Supplementary Fig. S3.

Table 3.

Metabolite levels in responders before and after rTMS treatment.

Characteristic	FEA-MDD	Post- rTMS	t	P	Cohen's d
Responders (n = 35)
GABA+ (L)	1.29 ± 0.17	1.41 ± 0.26	−2.11	0.041	0.541
Glx (L)	5.35 ± 0.69	5.08 ± 0.76	1.34	0.188	0.372
GABA+ (R)	1.54 ± 0.33	1.64 ± 0.23	−1.42	0.165	0.351
Glx (R)	5.72 ± 1.14	5.65 ± 1.18	0.27	0.784	0.060

Note: Data are presented as mean ± SD. n, number of participants; FEA-MDD, first episode adolescent major depressive disorder; post-rTMS, after 2 weeks of rTMS treatment; responders, the sum of the cure rate, markedly effectiveness rate, and effectiveness rate; L, left; R, right; Cohen’s d represents the standardized effect size for between-group differences.

Figure 3.

Metabolite levels in responders before and after rTMS Treatment. (A) After 2 weeks of rTMS treatment, GABA+ levels in the left DLPFC were significantly higher in the post-rTMS group than in the FEA-MDD group (P < 0.05). (B) No significant difference was observed in left DLPFC Glx levels between the two groups. (C–D) Neither GABA+ nor Glx levels in the right DLPFC showed statistically significant differences between the two groups. *FDR‐adjusted P  < 0.05. L, left; R, right.

Intra-group comparison of neurotransmitter levels between left and right DLPFC

In HCs, no significant interhemispheric differences were observed in GABA+ or Glx levels between the left and right DLPFC (GABA+: left, 1.48 ± 0.27 vs. right, 1.55 ± 0.37; Glx: left, 4.90 ± 1.04 vs. right, 4.45 ± 1.06; all P > 0.05).

In contrast, FEA-MDD showed significant interhemispheric differences: GABA+ levels were lower in the left DLPFC compared to the right (left: 1.30 ± 0.21 vs. right: 1.59 ± 0.34; P < 0.0001), while Glx levels were higher in the left DLPFC (left: 5.37 ± 0.69 vs. right: 5.83 ± 1.20; P = 0.034).

After rTMS treatment, interhemispheric differences persisted in the post-rTMS group: left DLPFC GABA+ levels remained lower than the right (1.37 ± 0.21 vs. 1.64 ± 0.27; P < 0.0001), and left DLPFC Glx levels were lower than the right (5.14 ± 0.72 vs. 5.67 ± 1.11; P = 0.012). Detailed statistics are presented in Supplementary Table S3 and Supplementary Fig. 4.

Correlation analysis

Pearson correlation analysis revealed a significant negative correlation between baseline HAMD scores and GABA+ levels in the left DLPFC voxel (r = −0.3433, P = 0.0435), indicating that greater depression severity was associated with lower GABA+ concentrations in the left DLPFC at baseline (Fig. 4). Additionally, Glx levels in the left DLPFC showed positive correlations with scores on the RRS, SDS, and SES. Detailed statistics are presented in Supplementary Table S4.

Figure 4.

Correlation between baseline neurotransmitter levels and clinical scale scores. (A) In the FEA-MDD group, left DLPFC GABA+ levels were negatively correlated with HAMD scores. (B–D) The left DLPFC Glx levels were positively correlated with scores on the RRS, SDS, and SES. HAMD, Hamilton Depression Rating Scale; RRS, Ruminative Responses Scale; SDS, Self-Rating Depression Scale; SES, Self-Esteem Scale.

Discussion

Our study highlights neurochemical dysregulation in the left DLPFC of FEA-MDD patients, characterized by reduced GABA+ levels and elevated Glx levels compared to HCs. These findings support the hypothesis that an imbalance in excitatory–inhibitory neurotransmission, particularly involving GABAergic and glutamatergic systems, plays a central role in the pathophysiology of MDD. Importantly, after rTMS treatment, we observed a partial normalization of these imbalances, with increased GABA+ levels and a trend toward decreased Glx levels in the left DLPFC. These neurochemical changes underscore the potential mechanisms underlying the antidepressant effects of rTMS and highlight its value as a non-invasive neuromodulatory intervention for adolescent depression.

Neurotransmitter asymmetry and its functional implications in adolescents with first-episode, drug-naïve MDD

One of the most striking findings in this study is the marked asymmetry in GABA+ and Glx levels between the left and right DLPFC in FEA-MDD. Specifically, our data showed that the left DLPFC exhibits significantly lower GABA+ and higher Glx levels compared to the right DLPFC in FEA-MDD, a pattern not observed in HCs. This observation aligns with the findings of previous studies in adult populations (Hasler et al., 2007; Kantrowitz et al., 2021), which suggest that the left DLPFC is particularly vulnerable to neurochemical dysregulation in MDD.

Our results should be interpreted considering the distinct neurodevelopmental stage of adolescence. The DLPFC of adolescents undergoes profound maturation, including synaptic pruning, increased myelination, and refinement of GABAergic inhibitory circuits characterized by dynamic changes in the balance between excitatory and inhibitory neurotransmission (Fuhrmann et al., 2015; Caballero et al., 2016; Caballero and Tseng, 2016). As a late-maturing region, the DLPFC is especially susceptible to disruption during this plastic window. The GABAergic deficits we observed in FEA-MDD may therefore represent a deviation or delay in this typical developmental trajectory, leading to an imbalance between excitation and inhibition that underpins the emotional and cognitive dysregulation characteristic of the disorder (Malinina et al., 2023).

GABA, the primary inhibitory neurotransmitter in the central nervous system, is crucial for regulating neuronal excitability and maintaining an appropriate balance within neural circuits. Dysregulated GABAergic transmission has long been implicated in the pathophysiology of FEA-MDD, with reduced GABA levels in key brain regions, including the prefrontal cortex, being consistently reported in both adult and adolescent populations (Dubin et al., 2016; Godfrey et al., 2018; Levitt et al., 2019). The lower GABA+ concentration in the left DLPFC observed in our study aligns with these findings and may reflect a disruption in the inhibitory control by this region over other brain areas involved in emotional processing and cognitive control. Notably, the left DLPFC plays a pivotal role in regulating mood, decision-making, and cognitive flexibility—functions often impaired in individuals with depression (Pizzagalli and Roberts, 2022; Taylor et al., 2022). Cutler et al. (2023) emphasized that GABA deficits in MDD impair the brain’s regulatory capacity, increasing emotional reactivity and susceptibility to stress (Su et al., 2020). This reduction in GABAergic signaling in the DLPFC may compromise prefrontal control over subcortical structures such as the amygdala, contributing to emotional dysregulation (Sanacora et al., 2012; Lin et al., 2022). Thus, GABA deficiencies may predispose adolescents to heightened anxiety, emotional overload, and impaired cognitive processing, all of which are hallmark features of MDD.

Concomitant with reduced in GABA+ levels, we observed elevated Glx concentrations in the left DLPFC of FEA-MDD patients. As the primary excitatory neurotransmitter, Glx has long been implicated in the pathophysiology of MDD. Alterations in glutamatergic transmission within key regions, such as the prefrontal cortex, hippocampus, and amygdala, have been reported in both animal and human studies (Sanacora et al., 2012; Kantrowitz et al., 2021). Elevated Glx levels indicate overactivity of the glutamatergic system in the DLPFC, which may contribute to excitotoxicity, synaptic dysfunction, and mood disturbances (Godfrey et al., 2021; Gonsalves et al., 2022). These findings align with patterns observed in adult treatment-resistant depression (TRD): Jiao et al. (2024) documented similar Glu elevations in adult TRD patients, suggesting that neurochemical alterations in adolescent MDD may share common features with adult forms of the disorder. The co-occurrence of low GABA+ and high Glx levels in the left DLPFC of adolescents with MDD points to a significant imbalance in neural circuits underlying mood regulation and cognitive function. This supports the hypothesis that depressive symptoms arise from an overactive excitatory system paired with insufficient inhibitory control.

While both the left and the right DLPFC are involved in higher-order cognitive functions such as working memory, attention, and emotion regulation (Li et al., 2017; Gavazzi et al., 2019), the left DLPFC is frequently associated with the regulation of positive emotions, cognitive control (Li et al., 2023), and executive functions, whereas the right DLPFC is more closely linked with processing negative emotions (Tran et al., 2023). In our study, we observed inconsistent GABA+ and Glx concentrations in the left and right DLPFC in adolescents with MDD, which was not the case in HCs. This aligns with research on brain lateralization in depression, which indicates that depression is often associated with increased activity in the right prefrontal cortex and decreased activity in the left prefrontal cortex, particularly during tasks involving emotion regulation (Horato et al., 2022; Yu et al., 2023). The neurochemical alterations observed in our study may reflect these functional asymmetries, with reduced GABAergic inhibition and increased glutamatergic excitation in the left DLPFC possibly contributing to the impaired emotional regulation that characterizes MDD.

Effects of rTMS on neurochemical imbalances

Another key finding in our study was that rTMS treatment significantly normalized GABA+ and Glx concentrations in the left DLPFC. Specifically, after 2 weeks of rTMS, GABA+ levels in the left DLPFC increased, while Glx levels decreased. These results align with previous studies (Dubin et al., 2016; Baeken et al., 2017; Levitt et al., 2019) and provide direct evidence for the neurochemical effects of rTMS in adolescent MDD. Our findings suggest that rTMS may offer a promising approach to restoring the balance between excitatory and inhibitory neurotransmission in the prefrontal cortex.

Subgroup analysis based on treatment response revealed differential neurochemical trajectories. Patients in the “cured” and “markedly effective” categories showed a trend toward increased GABA+ in the left DLPFC, consistent with the overall responder analysis. This aligns with prior studies suggesting that a successful antidepressant response is linked to enhanced GABAergic inhibition (Dubin et al., 2016; Levitt et al., 2019). Conversely, the ineffective subgroup displayed a slight reduction in GABA+, indicating a potential neurochemical resistance to rTMS, that may reflect a failure to restore inhibitory neurotransmission. Although these subgroup differences were not statistically significant, possibly due to limited sample size, they highlight the potential utility of baseline GABA+ levels as a predictor of rTMS response (Bhattacharyya et al., 2021; Godfrey et al., 2021). Future studies with larger samples are warranted to validate these preliminary findings and explore whether baseline Glx/GABA+ ratios could serve as a biomarker for stratifying treatment approaches.

rTMS has emerged as a promising non-invasive treatment for depression, but the mechanistic underpinnings of its effects on neurotransmitter concentrations remain incompletely understood. Several studies have demonstrated that rTMS can modulate local cortical excitability by stimulating neurons, enhancing neuroplasticity (Croarkin and MacMaster, 2019) and regulating neurotransmission in targeted areas, particularly in the left prefrontal cortex, a common target given its frequent hypoactivity in depression (Zheng et al., 2020). Augmenting GABAergic activity is crucial for restoring the equilibrium required for effective mood regulation and emotional control, thereby alleviating core symptoms of MDD (Levitt et al., 2019). In terms of neuroplasticity, rTMS promotes synaptic changes and enhances the connectivity and functioning of neural circuits related to emotion regulation (Krishnan et al., 2015). Mechanistically, it has been proposed that the effects of rTMS may share similarities with synaptic plasticity mechanisms such as long-term potentiation (LTP) (Cullen et al., 2019; MacMaster et al., 2019), potentially enhancing synaptic efficacy. These improvements in synaptic functionality may help normalize the aberrant activity patterns characteristic of depression, fostering resilience to stress and emotional dysregulation.

Moreover, the reduction in Glx levels observed in our study may reflect rTMS-induced downregulation of Glu release or increased uptake by astrocytes, which may reduce excitotoxic risk and stabilize the prefrontal–limbic circuits involved in mood regulation (Duman et al., 2019). Similar findings have been reported in previous studies, Michael et al. (2003) found that a single session of rTMS led to decreased levels of Glx in the DLPFC. The dual modulation of GABA+ and Glu underscores the role of rTMS in re-establishing excitatory–inhibitory balance and fostering a neurochemical environment conducive to emotional stability. However, the findings regarding Glx changes after rTMS are not entirely consistent. Some scholars have arrived at the opposite conclusion and found that DLPFC Glu levels increased after rTMS treatment (Yang et al., 2014). Dubin et al. (2016) were unable to find any significant difference in Glu or glutamate/glutamine peak Glx levels associated with rTMS for clinical depression.

Methodological factors (voxel placement, coil parameters, post-scan timing) undoubtedly influence MRS outcomes. Importantly, the adolescent brain exhibits heightened plasticity, a feature that may render it particularly responsive to neuromodulatory interventions such as rTMS (Fuhrmann et al., 2015; Croarkin and MacMaster, 2019). The significant increase in GABAergic activity we observed following treatment in our study may reflect this enhanced capacity for experience-dependent synaptic reorganization. Consequently, the neurobiological mechanisms underlying rTMS effects in adolescents may differ from those in adults, whose brains have reached a greater level of maturation and stability (Croarkin et al., 2021). Finally, the limited sample size (especially of non-responders) and the need for multiple comparisons also limit our ability to detect modest effects. In summary, the normalization of GABA+ by rTMS supports its role in restoring excitatory/inhibitory balance, but divergent Glx results across studies point to heterogeneity in patient characteristics, developmental stages, and study protocols (Conelea and Lieske, 2026).

Correlation analysis

We also observed a negative correlation between baseline left DLPFC GABA+ levels and HAMD scores, indicating that lower GABA+ concentrations may be associated with greater disease severity. Additionally, left DLPFC Glx levels were positively correlated with scores on the RRS, SDS, and SES. These findings suggest an excitatory/inhibitory imbalance in prefrontal cortex in these youths. This pattern aligns with a growing model of depression whereby deficient GABAergic inhibition leads to glutamatergic overactivity (Kantrowitz et al., 2021). For example, Kantrowitz et al. (2021) found that unmedicated MDD patients had elevated medial PFC Glx and lower GABA versus controls, with higher Glx paralleling greater depressive severity. In other words, higher Glx tends to mark more severe illness in depression, consistent with our correlation of Glx to higher SDS and RRS. The association of DLPFC Glx with rumination and depressive symptoms points to the role of glutamate in maladaptive thought patterns. The DLPFC is implicated in cognitive control and self-referential thinking; hyper-glutamatergic activity here may underlie perseverative negative focus (rumination), worsening mood. Notably, prior MRS work in adults has often reported reduced prefrontal Glx in medicated MDD patients (Bhattacharyya et al., 2021), but our data echo more recent findings in untreated samples: higher PFC Glx in severe, untreated depression (Kantrowitz et al., 2021). The positive correlation between Glx and SES scores was unexpected, as lower self-esteem is typically associated with greater depression severity. This finding may indicate a vulnerable compensatory mechanism in a specific patient subgroup, in which an initial glutamatergic surge within the DLPFC supports the cognitive regulation of self-worth (Arnsten, 2015; Duman et al., 2019). Such an observation aligns with the concept of "neuronal compensation," a phenomenon where individuals with better clinical outcomes or greater cognitive reserve exhibit enhanced neural resource mobilization when confronting brain-related challenges (Camprodon-Boadas et al., 2024). Thus, this elevation in Glx is probably not merely a marker of excitotoxic pathology, but rather reflects an active, potentially adaptive effort to sustain self-esteem. However, this observation remains purely hypothesis-generating, which underscores the need for longitudinal studies to clarify whether this putative compensatory state is transient or serves as a reliable predictor of long-term resilience.

Limitations of the study

However, several limitations of this study should be considered. First, the control group did not receive sham stimulation, precluding assessment of the psychological effects associated with rTMS treatment (e.g. placebo effects). Second, we could not rule out potential confounding effects of concurrent medications, as some adolescents with depression were taking psychotropic drugs throughout the study. Third, the lack of long-term follow-up data prevented us from drawing conclusions about the durability of the observed neurochemical changes after rTMS. Fourth, the DLPFC was localized using the "5 cm method" rather than individual MRI-based neuronavigation. While this is a common and practical clinical approach, it may introduce variability in stimulation targeting compared to more precise neuronavigation techniques. Finally, the relatively small sample size may have limited the generalizability of our findings. Future studies should address these limitations by incorporating larger, more diverse adolescent samples, examining additional brain regions, and including longitudinal follow-up assessments to confirm and extend our results. Integrating multimodal imaging and neurochemical profiling could further enhance our understanding of how rTMS modulates the neurocircuits implicated in MDD.

Conclusions

In summary, this study provides evidence of significant neurochemical dysregulation in the left DLPFC in FEA-MDD. rTMS may help treat adolescent depression by modulating neurotransmitter imbalances in the DLPFC. By increasing GABA+ and decreasing Glx levels, rTMS helps restore the balance necessary for mood regulation, offering a targeted approach to managing this complex disorder. Future research should continue to refine and expand on these findings to improve clinical applications.

Author contributions

Laiyang Ma (Conceptualization, Methodology, Writing – original draft, Writing—review & editing), Xiuyun Ma (Resources), Wenjing Huang (Data curation), Jun Wang (Validation), Songhong Yue (Resources), Lili Xu (Resources), Xiaojin Ma (Investigation), Yuhui Xiong (Supervision), Jing Zhang (Conceptualization, Writing—review & editing), Wanjun Hu (Visualization).

Conflicts of interest

The authors declare that there are no potential or actual conflicts of interest involving GE Healthcare or any other organizations/entities.